Responsive search ads (RSA) become the only option in search ad campaigns.

Last year, Google announced that they will depricate traditional extended text ads (ETA), shifting to the “responsive search ad” format only within search campaigns as of June 30, 2022. Now that the day is almost here, it appears we’re seeing a preview of it. From here forward, you will not have full control of what ad copy in search campaigns, or where each unit of ad copy is served within the ad.

The paid search community has made an assumption that even after extended text ads are eliminated, they will still be able to control their ad copy content and positioning by carefully manipulating the way they set up RSA. Perhaps because of this assumption, few ad professionals appear to be concerned about the shift to RSA-only ads. At this time, it is still possible to “mimic” an RSA if you carefully “pin” ad copy to Headline 1/2/3, and descriptions similarly.  But as it turns out, we now know this will not be allowed to work. Google is simply assigning a poor ad quality when this is done, preventing them from serving normally.

This is a seismic change in Google Ads. Google is seizing control of the presence of each unit of ad copy in each ad, as well as the sequence copy is delivered within the ad structure, perhaps down to the individual auction level. Instead of the advertiser making these decisions, Google’s automation will. Which could be a good thing, in theory – but we don’t know.

This change radically alters the control relationship between Google and search advertisers. Up until now, the ad copy served and what position each ad unit is served within the ads, have been determined by the advertiser.

There are numerous marketing, brand, performance, and regulatory considerations where the presence and position of ad headlines and descriptions better serve a business in a certain sequence. Ad delivery automation is not capable of “sentient reasoning” with these types of overarching factors.

The Advertising community may be in for a surprise when RSA-only fully rolls out on June 22nd

Now that the transition is here, advertisers and their agents are learning they will no longer be able to simulate ETA ads by selectively “pinning” ad copy. Advertisers and agencies lose much of their control over ad copy, for all ad auctions occurring in search campaigns. This would be agreeable IF it had been demonstrated that RSA’s outperform in operating margin (volume) and in return on ad spend (ROAS). But it has not.

Consider the fashion it’s been evaluated to date. The most important metric isn’t being used, and a baseline without RSA’s is absent. When an RSA is running within an ad group against ETA ads, we are not looking at a controlled, observable experiment.

Why might this concern the advertising community?

It is often said that there is resistance from the advertising community to the arrival of PPC automation. The premise is “Don’t be fearful of change…automation is good… you’ll get used to it”.

This line of reasoning with the advertising community misses the mark. For the most part, the advertising community understands that machine learning software innovations will not “replace” them. Most are technology enthusiasts and view software automation innovations positively.

What they are concerned with instead, is having visibility into their campaigns and the ability to maintain some control as technology proves itself. If the automation actually outperforms what advertisers have done manually, they will be enthusiastic about it. But there has been a lack of visibility and observability in RSA, and that is a good example of where the concerns about automation actually do originate.

Assessing RSA using operating margin and an ETA baseline

The operating margin for a PPC campaign is calculated using the same metrics as ROAS but calculated differently. ROAS is a ratio obtained by dividing conversion value by the ad spend. The operating margin metric is far more effective and is calculated by subtracting the ad spend amount from the conversion value. Its conversion value is adjusted for ad spend.  This is more important than ROAS to any advertiser because it’s the amount of money put into the top line by a campaign or account.

ROAS is more frequently the most important PPC metric because it is a normalized metric (a ratio) that can be understood immediately – even by non-technical managers! But it can also hide poor campaign management. For example, you can boast ROAS from 800 to 900% in a month while also bankrupting the advertiser because their conversion dollar volume may be contracting sharply. RSA performance for advertisers must be considered in light of operating margin (conversion dollar volume), or false conclusions may be reached about their performance.

Does RSA consume ETA conversion volume?

Google recommends always running one RSA and two ETA ads in each ad group and pushes advertisers hard for the early introduction of an RSA. Instead, Blastoff runs a pair of ETA-only ads to establish a baseline. Then we introduce an RSA once the ETA ads have enough clicks to reach the statistical power level to establish confidence in the numerical data.

Is the reason RSA’s look good because the RSA’s “consume” the ETA conversion, when brought into an ETA ad group?  Or are they actually superior? We would like to see experiments run showing operating margin comparisons for pairs of ETA ad groups in various sectors, with no RSA ads, then with an RSA ad in the ad group as recommended by Google.    I’ve never seen this topic addressed although it’s a measure of the quality of the RSA automation.

Are mandatory RSA’s fair to Google’s advertisers?

Most advertisers will be turning control of how their ad copy is served – the ad unit – over to Google’s black-box RSA automation algorithms.  This would be fine if it actually had been shown to improve results for advertisers.  But after working with RSA ads since they were introduced, we’re not convinced this is a good thing for advertisers and their service provider community.

If there were a truly viable competitor to Google, would they still be able to impose this radical change?

Thoughts or opinions?

Tweet us @BlastoffLabs on Twitter, or email us using the form on our website.